Update on "[kernel] Add template based unboxing"

Adding a new feature to allow users to bypass codegen and register their
kernels directly. This is very useful for custom kernels for custom ops.

Example usage:
```
Tensor& my_op(RuntimeContext& ctx, const Tensor& self, const Tensor& other, Tensor& out) {
    // ...
    return out;
}

Kernel my_kernel = Kernel.make_boxed_kernel("my_ns::my_op",EXECUTORCH_FN(my_op));

register_kernels({my_kernel});
```

Differential Revision: [D51553099](https://our.internmc.facebook.com/intern/diff/D51553099)

[ghstack-poisoned]

Author: larryliu0820

runtime/kernel/targets.bzl

@@ -30,7 +30,11 @@ def define_common_targets():
     runtime.cxx_library(
         name = "operator_registry_TWO_KERNELS_TEST_ONLY",
         srcs = ["operator_registry.cpp"],
-        exported_headers = ["operator_registry.h"],
+        exported_headers = [
+            "operator_registry.h",
+            "make_boxed_from_unboxed_functor.h",
+            "type_list.h",
+        ],
         visibility = [
             "//executorch/...",
             "@EXECUTORCH_CLIENTS",

runtime/kernel/test/make_boxed_from_unboxed_functor_test.cpp

@@ -18,13 +18,13 @@ using RuntimeContext = torch::executor::KernelRuntimeContext;
 using namespace torch::executor;
 
 Tensor& my_op_out(RuntimeContext& ctx, const Tensor& a, Tensor& out) {
-  (void) ctx;
-  (void) a;
+  (void)ctx;
+  (void)a;
   return out;
 }
 
 Tensor& set_1_out(RuntimeContext& ctx, Tensor& out) {
-  (void) ctx;
+  (void)ctx;
   out.mutable_data_ptr<int32_t>()[0] = 1;
   return out;
 }